Laser retina imaging technology enables creation of a clear picture in any retina area. It is focus-free, i.e. irrelevant of refractive disorder. So in many cases of low vision, one can see the picture by one's own eyesight. In this article we describe possible applications of this emerging technology, including 1. Development of eyesight-aid for low vision, an HMD (headmounted display) on which this technology is implemented, 2. Creating a vision sensitive area map of retina, a novel notion, precisely. Especially this may innovate perimetry technologies in ophthalmology important in finding disorders in vision. We list up requirements for the eyesight-aid, including a compensation of partial visual field loss. Finally we indicate that part of our requirements is feasible by presenting our preliminary experiments. Indeed, we show that it is possible to detect a pictogram (“Exit”) and pedestrian boundaries on the road-side both in the view by existing pattern recognition programs (SURF method).

One of the problems regarding graphics for blind individual is that we have few data on how he (she) acquires graphical information through touching. In this article, we propose a method of tracking the touch point of his (her) fingertip on graphics material using cameras (an analog of eye-tracking) for analyzing the image acquisition process. We exhibit the method on an edge-outlined tactile picture of a flower and show that it yields basic features of the original outlined flower by an experiment relying on a visually impaired subject. As a linked yet remotely material in Section 3.1 we describe our other application of online tactile map in which we have introduced the idea of sharing positional information between touching a tactile picture and looking into a graphical picture using a tracking technology.

We describe a method of sharing information between a tactile map on a paper and an online map on the information terminals (e.g., Tablet computers such as iPad). We use "digital pen" technology which is used as an input method of hand-written characters (e.g., Pegasus technologies such as Note Taker). Our method is based on synchronizing the location of contact point on the tactile-map with the corresponding point on the original online map achieved through the following two steps: 1) Create a road-only tactile map: Extract road pictures from the target online map and create a simple road-only map in SVG (scalable vector graphics), then copy it onto a tactile paper. 2) Point on the tactile map by digital pen. Then, the location (coordinates) of the contact point is synchronized by the digital pen synchronization mechanism and mapped onto the corresponding location in the online map (a calibration is done beforehand between the maps using SVG facility). Thus, various information of the online-map can be provided to the visually impaired user in the text form from the online-map database. Specific information pertaining to the visually impairness such as entrance of the buildings could be also supplied from the self-maintained database.

Creating a virtual reality (VR) by combining visual information and pseudo touch sensing is studied actively. However, being strictly detached from visual information, it is not easy for visually impaired individuals to sense a visual VR world. The purpose of our study is the development of, so we call, haptic VR technologies for aiding visually impaired individuals sense his (her) surroundings and recognize geometric constructions in the space. Our haptic VR system for the cane walk is installed as follows: in the VR world we put (draw) a cane and a hand with its 5 fingers holding the cane. In real word, we wear a data glove on one hand and this hand is superposed on the virtual hand (this is achieved by each finger being force provided through the data-glove). Thus our VR sensing is created by the pseudo haptic sensing at the fingers for which the feedback forces are provided by the data-glove worn on the hand. In our haptic VR space of cane walk, we have prepared "studded paving blocks" to aid the visually impaired for the pavement, a "cane (white cane)" and a "hand" to grasp and manipulate the cane on the virtual pavement. An actual hand wearing the data-glove on it is superposed on the virtual hand by the system. Thus the touch and investigation on the virtual paving blocks by the cane in the haptic VR space is conveyed to the actual hand via the data-glove. We also show some experimental vibration value and haptic value data obtained at the index finger of the cane grasping hand during the real walking on studded paving blocks and some skin surface voltage of electromyogram (EMG) data.

We have built an arm-navigation assisting system for a visually impaired person (user) to reach an object on the table, where optical tracking of marks attacked both on the objects and on his arm is used in order to augment his sight. The system helps him with giving spacial information of the workspace so that he creates a cognitive map of the workspace. For this purpose degrees of congestion on the workspace must be conveyed to the user. Starting from the description of the assisting system, we propose in this paper a method of judging the degrees of congestion of the workspace around arm. There are five of them: from “narrow” to “broad,” which are determined by using well-established Neural Network techniques on the basis of the spacial data obtained by the Distance Field Model (DFM) representation of the workspace. Defining spaciousness by entropy-like measure based on the DFM data is also proposed separately.

Motivated from the result of our field work study that low vision people are hardly noticing public signs on the streets and in the interior of buildings even under clear weather, in this report we continue our research on application of eye tracking technology for low vision aids. We start from a short characterization of low vision from viewing standpoint and show that low vision person can basically recognize target object by his residual sight on his mobile display if we send an enlarged clear vision of the target. Taking advantage of this possible enhancement of low vision we explore eye tracking technology for helping him with navigation during this walking time. We show that classical scanpath technique for localizing regions of interest (ROIs) is applicable with low vision as well. Then we proceed to examine possible enhancement of low vision by (1) segmenting out public signs from his ROI, and (2) sending its enhanced vision back to his mobile monitor. We also show a preliminary result of public sign recognition in the view by using a fast pattern matching technique called “boosting,” liking to a future system of vision navigator for guiding the gaze of low vision to a missing public sign and zooming into it. Optimization of classifier programs is discussed from decision tree standpoint separately.

We describe a HAptic Walk-Guide simulator (HAWG) which gives a simulator-training of an indoor pathway finding for a visually impaired person through haptic recognition. The purpose of the walk-guide simulator is to present user environmental information along the path by haptic means in order to help him create a cognitive map of the environment. In this report the emphasis is placed on the following extension of the system implemented in the latest version: 1. When encountered with an object, verbal description of the object is given so that he can know his current location and position; 2. To help efficient training various features are implemented including: teacher-student (master-slave) mode for dual haptic devices and moving in the layout by inputting arrow keys. The latter facility makes the user record the moves in a file; later, it can be recovered from the file so that he can repeat the training; 3. From one layout (say, from a hallway) the user can transit into another layout (into a room) continually. We have also incorporated a real-time tactile graphic display in HAWG in order to augment the possible drawback of extreme locality of the haptic sensing.

For a visually impaired person identifying the landmark he is expecting in his cognitive map is crucial. To let him do this, we approach as follows: we set landmarks in the environment by RFIDs and then we identify, trace and localize them. In the first part we describe our approach in detail, including landmarking by passive RFIDs and tracing them by autonomous robots. As visually impairedness badly interfere with the detection of changes in the environment, in the second part, we describe localization of active RFIDs which are deployed in the room and subject to spontaneous move. Our approach for this problem is: 1. place several anchoring (fixed) RFIDs, 2. for the estimation of a possible location of movable RFIDs we use clustering by SOM, 3. we navigate robots iterating the estimation algorithm until we reach the desired object (RFID).

The construction of information ensured environment for the visually impaired (say, in the college life) combining data carrier and network technologies includes the problem of obtaining and conveying environmental information. As the visually impaired have keen difficulties in sensing changes of surroundings, in order to assist them we usually need an interface (say, a guide dog) between people and the environment. We put RFID tags on surrounding objects so that we can treat them as landmarks of the environment. Then we can design autonomous robots to sense the changes of locations of objects and situations. In this paper we describe a general navigation algorithm of such robots based on distance information obtained, say, by super sonic sensors which are equipped on the robot. Our algorithm is designed on the so-called distance field space model (DFM), which can be considered as a natural description of objects and space on the basis of a fuzzy set theory. Though major part of this paper is devoted to the development of the navigation principle based on DFM, our RFID-sensing robots are not equipped with this algorithm yet.

Data carrier technology such as barcode and RFID greatly improves information ensuring for the visually impaired; using the key kept in the data carrier attached to an equipment or surrounding objects, one can obtain detailed information from a database on the network. The purpose of the project we report here is the construction of such an information ensured environment in the college life. In the first part we present two concrete examples: getting a voice access to announcements on the bulletin board by using barcode, and messaging system between students and teacher by using RFID tags. In the second part, after giving the idea of setting marks in the environment by using RFID tags and tracing their moves, we present an RFID-sensing robot which cruises in the environment and checks changes occurred in it by using the record in RFID tags; by this we can cope with the dynamic changes of the environment for information ensuring for the visually impaired.

An iterative algorithm for generating all partitions of the set {1,…,n} is presented. An empirical test shows that the new algorithm is faster than the previously fastest algorithm recently proposed by Er on some computers, though the former is slower than the latter on a computer where fast recursive call is provided based on an RISC architecture.

Two variable selection criteria are proposed for converting a decision table to a near-optimum decision tree in the sense of minimal average cost of testing. A criterion, Q, is introduced that is based on the potential of a decision table. The previously known criterion 'loss' and Q are combined into a third criterion O. The performance of the three criteria is examined both theoretically and experimentally. Of most importance is that Q and O do not select a nonessential variable, while 'loss' may do so. It is also shown that the performance of the three criteria is not worse than that of any other known heuristics, at least for a particular example. The algorithm requires at most O(L/sup 2/2/sup L/) operations, where L is the arity of an input table.>

On the basis of a backtrack procedure for lexicographic enumeration of all subsets of a set of n elements, we give an algorithm both for determining all bases consisting of functions from a given complete set in a considered subset of the set of k-valued logical functions, and for enumeration of all classes of bases in the subset. We use the lexicographic algorithm also for solving knapsack and minimal covering problems. A cut technique is described which is used in these algorithms to reduce the number of examined subsets of {1, …, n}.

The construction problems of optimum sequence trees (or digital search trees) are considered in the following frameworks: 1. construction of optimum trees from a set of keys, 2. transformation of an arbitrary tree into an optimum one, 3. optimum insertions of keys into an optimum tree, 4. optimum deletions of keys from an optimum tree. Algorithms are shown. The number of operations needed for the algorithm in the framework 1 is at most $O(N^{2}L)$, and in the framework 2 it is at most $O(N^{3}L)$ both with $O(N)$ storage locations, where N and L are the number of keys and the length of coded keys respectively. Necessary and sufficient conditions for the optimality of sequence trees are also given.